Since these elements now use `white-space: pre-wrap` since
784665d4ce, it's fine to use newlines
for formatting, which is smaller and a bit less complicated.
Rollup of 8 pull requests
Successful merges:
- #100599 (Add compiler error E0523 long description and test)
- #107471 (rustdoc: do not include empty default-settings tag in HTML)
- #107555 (Modify existing bounds if they exist)
- #107662 (Turn projections into copies in CopyProp.)
- #107695 (Add test for Future inflating arg size to 3x )
- #107700 (Run the tools builder on all PRs)
- #107706 (Mark 'atomic_mut_ptr' methods const)
- #107709 (Fix problem noticed in PR106859 with char -> u8 suggestion)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Fix problem noticed in PR106859 with char -> u8 suggestion
HN reader `@ayosec` noticed that my #106859 a few weeks back, malfunctions if you have a Unicode escape, the code suggested b'\u{0}' if you tried to use '\u{0}' where a byte should be, when of course b'\u{0}' is not a byte literal, regardless of the codepoint you can't write Unicode escapes in a byte literal at all.
My proposed fix here just checks that the "character" you wrote is fewer than 5 bytes, thus allowing \x7F and similar escapes but conveniently forbidding even the smallest Unicode escape \u{0} before offering the suggestion as before.
I have provided an updated test which includes examples which do and don't work because of this additional rule.
Turn projections into copies in CopyProp.
The current implementation can leave behind projections that are moved out several times.
This PR widens the check to turn such moves into copies: a move out of a projection of a copy is equivalent to a copy of the original projection.
Modify existing bounds if they exist
Fixes#107335.
This implementation is kinda gross but I don't really see a better way to do it.
This primarily does two things: Modifies `suggest_constraining_type_param` to accept a new parameter that indicates a span to be replaced instead of added, if presented, and limit the additive suggestions to either suggest a new bound on an existing bound (see newly added unit test) or add the generics argument if a generics argument wasn't found.
The former change is required to retain the capability to add an entirely new bounds if it was entirely omitted.
r? ``@compiler-errors``
Fix suggestions rendering when the diff span is multiline
Fixes#92741
cc `@estebank`
I think, I finally fixed. I still want to go back and try to clean up the code a bit. I'm open to suggestions.
Some examples of the new suggestions:
```
help: consider removing the borrow
|
2 - &
|
```
```
help: consider removing the borrow
|
2 - &
3 - mut
|
```
```
help: consider removing the borrow
|
2 - &
3 - mut if true { true } else { false }
2 + if true { true } else { false }
|
```
Should we add a test to ensure this behavior doesn't disappear in the future?
There is a distinction between running this on wasm and i686, even though they should be
identical. This technically is not _incorrect_, it's just an unexpected difference, which is
worth investigating, but not for correctness.
Since we're changing a bunch of stuff, necessary to remove some codegen tests
which look for specific things. Also attempting to restart a test which timed out, maybe due to
fastly failing?
This adds one more test that should track improvements to generator
layout, like #62958 and #62575.
In particular, this test highlights suboptimal layout, as the storage
for the argument future is not being reused across its usage as `upvar`,
`local` and `awaitee` (being polled to completion).
make &mut !Unpin not dereferenceable, and Box<!Unpin> not noalias
See https://github.com/rust-lang/unsafe-code-guidelines/issues/381 and [this LLVM discussion](https://discourse.llvm.org/t/interaction-of-noalias-and-dereferenceable/66979). The exact semantics of how `noalias` and `dereferenceable` interact are unclear, and `@comex` found a case of LLVM actually exploiting that ambiguity for optimizations. I think for now we should treat LLVM `dereferenceable` as implying a "fake read" to happen immediately at the top of the function (standing in for the spurious reads that LLVM might introduce), and that fake read is subject to all the usual `noalias` restrictions. This means we cannot put `dereferenceable` on `&mut !Unpin` references as those references can alias with other references that are being read and written inside the function (e.g. for self-referential generators), meaning the fake read introduces aliasing conflicts with those other accesses.
For `&` this is already not a problem due to https://github.com/rust-lang/rust/pull/98017 which removed the `dereferenceable` attribute for other reasons.
Regular `&mut Unpin` references are unaffected, so I hope the impact of this is going to be tiny.
The first commit does some refactoring of the `PointerKind` enum since I found the old code very confusing each time I had to touch it. It doesn't change behavior.
Fixes https://github.com/rust-lang/miri/issues/2714
EDIT: Turns out our `Box<!Unpin>` treatment was incorrect, too, so the PR also fixes that now (in codegen and Miri): we do not put `noalias` on these boxes any more.
Sort Generator `print-type-sizes` according to their yield points
Especially when trying to diagnose runaway future sizes, it might be more intuitive to sort the variants according to the control flow (aka their yield points) rather than the size of the variants.
Refine error spans for "The trait bound `T: Trait` is not satisfied" when passing literal structs/tuples
This PR adds a new heuristic which refines the error span reported for "`T: Trait` is not satisfied" errors, by "drilling down" into individual fields of structs/enums/tuples to point to the "problematic" value.
Here's a self-contained example of the difference in error span:
```rs
struct Burrito<Filling> {
filling: Filling,
}
impl <Filling: Delicious> Delicious for Burrito<Filling> {}
fn eat_delicious_food<Food: Delicious>(food: Food) {}
fn will_type_error() {
eat_delicious_food(Burrito { filling: Kale });
// ^~~~~~~~~~~~~~~~~~~~~~~~~ (before) The trait bound `Kale: Delicious` is not satisfied
// ^~~~ (after) The trait bound `Kale: Delicious` is not satisfied
}
```
(kale is fine, this is just a silly food-based example)
Before this PR, the error span is identified as the entire argument to the generic function `eat_delicious_food`. However, since only `Kale` is the "problematic" part, we can point at it specifically. In particular, the primary error message itself mentions the missing `Kale: Delicious` trait bound, so it's much clearer if this part is called out explicitly.
---
The _existing_ heuristic tries to label the right function argument in `point_at_arg_if_possible`. It goes something like this:
- Look at the broken base trait `Food: Delicious` and find which generics it mentions (in this case, only `Food`)
- Look at the parameter type definitions and find which of them mention `Filling` (in this case, only `food`)
- If there is exactly one relevant parameter, label the corresponding argument with the error span, instead of the entire call
This PR extends this heuristic by further refining the resulting expression span in the new `point_at_specific_expr_if_possible` function. For each `impl` in the (broken) chain, we apply the following strategy:
The strategy to determine this span involves connecting information about our generic `impl`
with information about our (struct) type and the (struct) literal expression:
- Find the `impl` (`impl <Filling: Delicious> Delicious for Burrito<Filling>`)
that links our obligation (`Kale: Delicious`) with the parent obligation (`Burrito<Kale>: Delicious`)
- Find the "original" predicate constraint in the impl (`Filling: Delicious`) which produced our obligation.
- Find all of the generics that are mentioned in the predicate (`Filling`).
- Examine the `Self` type in the `impl`, and see which of its type argument(s) mention any of those generics.
- Examing the definition for the `Self` type, and identify (for each of its variants) if there's a unique field
which uses those generic arguments.
- If there is a unique field mentioning the "blameable" arguments, use that field for the error span.
Before we do any of this logic, we recursively call `point_at_specific_expr_if_possible` on the parent
obligation. Hence we refine the `expr` "outwards-in" and bail at the first kind of expression/impl we don't recognize.
This function returns a `Result<&Expr, &Expr>` - either way, it returns the `Expr` whose span should be
reported as an error. If it is `Ok`, then it means it refined successfull. If it is `Err`, then it may be
only a partial success - but it cannot be refined even further.
---
I added a new test file which exercises this new behavior. A few existing tests were affected, since their error spans are now different. In one case, this leads to a different code suggestion for the autofix - although the new suggestion isn't _wrong_, it is different from what used to be.
This change doesn't create any new errors or remove any existing ones, it just adjusts the spans where they're presented.
---
Some considerations: right now, this check occurs in addition to some similar logic in `adjust_fulfillment_error_for_expr_obligation` function, which tidies up various kinds of error spans (not just trait-fulfillment error). It's possible that this new code would be better integrated into that function (or another one) - but I haven't looked into this yet.
Although this code only occurs when there's a type error, it's definitely not as efficient as possible. In particular, there are definitely some cases where it degrades to quadratic performance (e.g. for a trait `impl` with 100+ generic parameters or 100 levels deep nesting of generic types). I'm not sure if these are realistic enough to worry about optimizing yet.
There's also still a lot of repetition in some of the logic, where the behavior for different types (namely, `struct` vs `enum` variant) is _similar_ but not the same.
---
I think the biggest win here is better targeting for tuples; in particular, if you're using tuples + traits to express variadic-like functions, the compiler can't tell you which part of a tuple has the wrong type, since the span will cover the entire argument. This change allows the individual field in the tuple to be highlighted, as in this example:
```
// NEW
LL | want(Wrapper { value: (3, q) });
| ---- ^ the trait `T3` is not implemented for `Q`
// OLD
LL | want(Wrapper { value: (3, q) });
| ---- ^~~~~~~~~~~~~~~~~~~~~~~~~ the trait `T3` is not implemented for `Q`
```
Especially with large tuples, the existing error spans are not very effective at quickly narrowing down the source of the problem.
Rollup of 5 pull requests
Successful merges:
- #107553 (Suggest std::ptr::null if literal 0 is given to a raw pointer function argument)
- #107580 (Recover from lifetimes with default lifetimes in generic args)
- #107669 (rustdoc: combine duplicate rules in ayu CSS)
- #107685 (Suggest adding a return type for async functions)
- #107687 (Adapt SROA MIR opt for aggregated MIR)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Adapt SROA MIR opt for aggregated MIR
The pass was broken by https://github.com/rust-lang/rust/pull/107267.
This PR extends it to replace:
```
x = Struct { 0: a, 1: b }
y = move? x
```
by assignment between locals
```
x_0 = a
x_1 = b
y_0 = move? x_0
y_1 = move? x_1
```
The improved pass runs to fixpoint, so we can flatten nested field accesses.
Suggest std::ptr::null if literal 0 is given to a raw pointer function argument
Implementation feels a little sus (we're parsing the span for a `0`) but it seems to fall in line the string-expected-found-char condition right above this check, so I think it's fine.
Feedback appreciated on help text? I think it's consistent but it does sound a little awkward maybe?
Fixes#107517
Adds the extended error documentation for E0523 to indicate that the
error is no longer produced by the compiler.
Update the E0464 documentation to include example code that produces the
error.
Remove the error message E0523 from the compiler and replace it with an
internal compiler error.
The code that consumes PointerKind (`adjust_for_rust_scalar` in rustc_ty_utils)
ended up using PointerKind variants to talk about Rust reference types (& and
&mut) anyway, making the old code structure quite confusing: one always had to
keep in mind which PointerKind corresponds to which type. So this changes
PointerKind to directly reflect the type.
This does not change behavior.
don't point at nonexisting code beyond EOF when warning about delims
Previously we would show this:
```
warning: unnecessary braces around block return value
--> /tmp/bad.rs:1:8
|
1 | fn a(){{{
| ^ ^
|
= note: `#[warn(unused_braces)]` on by default
help: remove these braces
|
1 - fn a(){{{
1 + fn a(){{
|
```
which is now hidden in this case.
We would create a span spanning between the pair of redundant {}s but there is only EOF instead of the `}` so we would previously point at nothing. This would cause the debug assertion ice to trigger. I would have loved to just only point at the second delim and say "you can remove that" but I'm not sure how to do that without refactoring the entire diagnostic which seems tricky. :( But given that this does not seem to regress any other tests we have, I think this edge-casey enough be acceptable.
Fixes https://github.com/rust-lang/rust/issues/107423
r? `@compiler-errors`
Especially when trying to diagnose runaway future sizes, it might be
more intuitive to sort the variants according to the control flow
(aka their yield points) rather than the size of the variants.
Previously we would show this:
```
warning: unnecessary braces around block return value
--> /tmp/bad.rs:1:8
|
1 | fn a(){{{
| ^ ^
|
= note: `#[warn(unused_braces)]` on by default
help: remove these braces
|
1 - fn a(){{{
1 + fn a(){{
|
```
which is now hidden in this case.
We would create a span spanning between the pair of redundant {}s but there is only EOF instead of the `}` so we would previously point at nothing.
This would cause the debug assertion ice to trigger.
I would have loved to just only point at the second delim and say "you can remove that" but I'm not sure how to do that without refactoring the entire diagnostic which seems tricky. :(
But given that this does not seem to regress any other tests we have, I think this edge-casey enough be acceptable.
Fixes https://github.com/rust-lang/rust/issues/107423
r? @compiler-errors
Replace nbsp in all rustdoc code blocks
Based on #106125 by `@dtolnay` — this PR fixes the line wrapping bug.
Fixes#106098. This makes code copyable from rustdoc rendered documentation into a Rust source file.
Fix suggestion for coercing Option<&String> to Option<&str>
Fixes#107604
This also makes the diagnostic `MachineApplicable`, and runs `rustfix` to check we're not producing incorrect code.
``@rustbot`` label +A-diagnostics
Don't cause a cycle when formatting query description that references a FnDef
When a function returns `-> _`, we use typeck to compute what the resulting type of the body _should_ be. If we call another query inside of typeck and hit a cycle error, we attempt to report the cycle error which requires us to compute all of the query descriptions for the stack.
However, if one of the queries in that cycle has a query description that references this function as a FnDef type, we'll cause a *second* cycle error from within the cycle error reporting code, since rendering a FnDef requires us to compute its signature. This causes an unwrap to ICE, since during the *second* cycle reporting code, we try to look for a job that isn't in the active jobs list.
We can avoid this by using `with_no_queries!` when computing these query descriptions.
Fixes#107089
The only drawback is that the rendering of opaque types in cycles regresses a bit :| I'm open to alternate suggestions about how we may handle this...
Emit warnings on unused parens in index expressions
Fixes: #96606.
I am not sure what the best term for "index expression" is. Is there a better term we could use?
Don't generate unecessary `&&self.field` in deriving Debug
Since unsized fields may only be the last one in a struct, we only need to generate a double reference (`&&self.field`) for the final one.
cc `@nnethercote`
Suggest `{var:?}` when finding `{?:var}` in inline format strings
Link to issue: https://github.com/rust-lang/rust/issues/106572
This is my first PR to this project, so hopefully I can get some good pointers with me from the first PR.
Currently my idea was to test out whether or not this is the correct solution to this issue and then hopefully expand upon the idea to not only work for Debug formatting but for all of them. If this is a valid solution, I will create a new issue to give a better error message to a broader range of wrong-order formatting.
Erase regions before doing uninhabited check in borrowck
~Also, fingerprint query keys/values when debug assertions are enabled. This should make it easier to check for issues like this without `-Cincremental`, and make UI tests a bit cleaner.~ edit: moving that to a separate PR
Fixes#107505
Remove both StorageLive and StorageDead in CopyProp.
Fixes https://github.com/rust-lang/rust/issues/107511https://github.com/rust-lang/rust/pull/106908 removed StorageDead without the accompanying StorageLive. In loops, execution would see repeated StorageLive, without any StorageDead, which is UB.
So when removing storage statements, we have to remove both StorageLive and StorageDead.
~I also added a MIR validation pass for StorageLive. It may be a bit overzealous.~
Improve diagnostic for missing space in range pattern
Improves the diagnostic in #107425 by turning it into a note explaining the parsing issue.
r? `@compiler-errors`
Revert "Teach parser to understand fake anonymous enum syntax" and related commits
anonymous enum types are currently ambiguous in positions like:
* `|` operator: `a as fn() -> B | C`
* closure args: `|_: as fn() -> A | B`
I first tried to thread around `RecoverAnonEnum` into all these positions, but the resulting complexity in the compiler is IMO not worth it, or at least worth a bit more thinking time. In the mean time, let's revert this syntax for now, so we can go back to the drawing board.
Fixes#107461
cc: `@estebank` `@cjgillot` #106960
---
### Squashed revert commits:
Revert "review comment: Remove AST AnonTy"
This reverts commit 020cca8d36.
Revert "Ensure macros are not affected"
This reverts commit 12d18e4031.
Revert "Emit fewer errors on patterns with possible type ascription"
This reverts commit c847a01a3b.
Revert "Teach parser to understand fake anonymous enum syntax"
This reverts commit 2d82420665.
Revert "review comment: Remove AST AnonTy"
This reverts commit 020cca8d36.
Revert "Ensure macros are not affected"
This reverts commit 12d18e4031.
Revert "Emit fewer errors on patterns with possible type ascription"
This reverts commit c847a01a3b.
Revert "Teach parser to understand fake anonymous enum syntax"
This reverts commit 2d82420665.
Add proc-macro boilerplate to crt-static test
I was seeing this failure when running ui tests with with a `-Cpanic=abort` stdlib targeting fuchsia:
```
---- [ui] tests/ui/proc-macro/crt-static.rs stdout ----
normalized stderr:
warning: building proc macro crate with `panic=abort` may crash the compiler should the proc-macro panic
warning: 1 warning emitted
The actual stderr differed from the expected stderr.
```
`force-host` was enough to stop it from running/failing, not sure if I should also add `needs-unwind`?
Fix syntax in `-Zunpretty-expanded` output for derived `PartialEq`.
If you do `derive(PartialEq)` on a packed struct, the output shown by `-Zunpretty=expanded` includes expressions like this:
```
{ self.x } == { other.x }
```
This is invalid syntax. This doesn't break compilation, because the AST nodes are constructed within the compiler. But it does mean anyone using `-Zunpretty=expanded` output as a guide for hand-written impls could get a nasty surprise.
This commit fixes things by instead using this form:
```
({ self.x }) == ({ other.x })
```
r? ``@RalfJung``
Remove confusing 'while checking' note from opaque future type mismatches
Maybe I'm just misinterpreting the wording of the note. The only value I can see in this note is that it points out where the async's opaque future is coming from, but the way it's doing it is misleading IMO.
For example:
```rust
note: while checking the return type of the `async fn`
--> $DIR/dont-suggest-missing-await.rs:7:24
|
LL | async fn make_u32() -> u32 {
| ^^^ checked the `Output` of this `async fn`, found opaque type
```
We point at the type `u32` in the HIR, but then say "found opaque type". We also say "while checking"... but we're typechecking a totally different function when we get this type mismatch!
r? ``@estebank`` but feel free to reassign and/or take your time reviewing this. I'd be inclined to also discuss reworking the presentation of this type mismatch to restore some of these labels in a way that makes it more clear what it's trying to point out.
The check previously matched this, and suggested adding a missing
`struct`:
pub Foo(...):
It was probably intended to match this instead (semicolon instead of
colon):
pub Foo(...);
Strengthen validation of FFI attributes
Previously, `codegen_attrs` validated the attributes `#[ffi_pure]`, `#[ffi_const]`, and `#[ffi_returns_twice]` to make sure that they were only used on foreign functions. However, this validation was insufficient in two ways:
1. `codegen_attrs` only sees items for which code must be generated, so it was unable to raise errors when the attribute was incorrectly applied to macros and the like.
2. the validation code only checked that the item with the attr was foreign, but not that it was a foreign function, allowing these attributes to be applied to foreign statics as well.
This PR moves the validation to `check_attr`, which sees all items. It additionally changes the validation to ensure that the attribute's target is `Target::ForeignFunction`, only allowing the attributes on foreign functions and not foreign statics. Because these attributes are unstable, there is no risk for backwards compatibility. The changes also ending up making the code much easier to read.
This PR is best reviewed commit by commit. Additionally, I was considering moving the tests to the `attribute` subdirectory, to get them out of the general UI directory. I could do that as part of this PR or a follow-up, as the reviewer prefers.
CC: #58328, #58329
Do not depend on Generator trait when deducing closure signature
1. Do not depend on `Generator` trait when deducing closure signature.
2. Compare the name of the `Generator::Return` associated item, rather than its order in the trait. Seems more stable this way.
Fixing confusion between mod and remainder
Like many programming languages, rust too confuses remainder and modulus. The `%` operator and the associated `Rem` trait is (as the trait name suggests) the remainder, but since most people are linguistically more familiar with the modulus the documentation sometimes claims otherwise. This PR tries to fix this problem in rustc.
If you do `derive(PartialEq)` on a packed struct, the output shown by
`-Zunpretty=expanded` includes expressions like this:
```
{ self.x } == { other.x }
```
This is invalid syntax. This doesn't break compilation, because the AST
nodes are constructed within the compiler. But it does mean anyone using
`-Zunpretty=expanded` output as a guide for hand-written impls could get
a nasty surprise.
This commit fixes things by instead using this form:
```
({ self.x }) == ({ other.x })
```
Modify primary span label for E0308
Looking at the reactions to https://hachyderm.io/`@ekuber/109622160673605438,` a lot of people seem to have trouble understanding the current output, where the primary span label on type errors talks about the specific types that diverged, but these can be deeply nested type parameters. Because of that we could see "expected i32, found u32" in the label while the note said "expected Vec<i32>, found Vec<u32>". This understandably confuses people. I believe that once people learn to read these errors it starts to make more sense, but this PR changes the output to be more in line with what people might expect, without sacrificing terseness.
Fix#68220.
incremental: migrate diagnostics
- Apply the diagnostic migration lints to more functions on `Session`, namely: `span_warn`, `span_warn_with_code`, `warn` `note_without_error`, `span_note_without_error`, `struct_note_without_error`.
- Add impls of `IntoDiagnosticArg` for `std::io::Error`, `std::path::Path` and `std::path::PathBuf`.
- Migrate the `rustc_incremental` crate's diagnostics to translatable diagnostic structs.
r? `@compiler-errors`
cc #100717
Use `ObligationCtxt::new_in_snapshot` in `satisfied_from_param_env`
We can evaluate nested `ConstEvaluatable` obligations in an evaluation probe, which will ICE if we use `ObligationCtxt::new`.
Fixes#107474Fixes#106666
r? `@BoxyUwU` but feel free to reassign
cc `@JulianKnodt` who i think added this assertion code
Not sure if the rustdoc test is needed, but can't hurt. They're the same root cause, though.
Implement unsizing in the new trait solver
This makes hello world compile! Ignore the first commit, that's just #107146 which is waiting on merge.
I'll leave some comments inline about design choices that might be debatable.
r? `@lcnr` (until we have a new trait solver reviewer group...)
Fix invalid float literal suggestions when recovering an integer
Only suggest adding a zero to integers with a preceding dot when the change will result in a valid floating point literal.
For example, `.0x0` should not be turned into `0.0x0`.
r? nnethercote
Only suggest adding a zero to integers with a preceding dot when the change will
result in a valid floating point literal.
For example, `.0x0` should not be turned into `0.0x0`.
